tim   .


newpad.gif (50 bytes)

.   hunkin





The life of the workers picking the products at an Amazon warehouse is uncannily like a video game. The barcode reading ‘picking guns’ direct the pickers to the next product and tell them the length of time they have to reach it. I’ve read there’s even a rising beep as they near their allotted time. Fun as a game for a couple of minutes but hell to do all day, every day. Simply going the toilet is difficult because the warehouses are so large it’s usually a very long walk. Each time you leave the warehouse you have to pass through a full body scanner. The canteen is usually so far away that the scan and the walk takes over half the allotted 30 minute meal break. An ex Amazon engineer I met told me the story of a picker so terminally pissed off that he dropped his picking gun down the toilet and shat on it.

There’ve been newspaper articles and TV programs about the inhuman working conditions and Youtube has lots of videos by disgruntled employees. I first read the details in a book called ‘Hired’ by a UK journalist called James Bloodworth. He worked as a picker at Amazon for three months, followed by spells at an insurance company call center, at a Blackpool care home and as a Uber driver. He concluded that although the ‘freedom’ of the gig economy – working whatever hours you chose etc – is seductive, there is actually much less freedom than it seems. Uber drivers are not allowed to refuse rides allocated to them, zero hours workers who are not allowed to take on secondary jobs even if there is no work….and of course the Amazon pickers who have their moves between products precisely timed. Bloodworth particularly lamented the loss of the union movements’ hard won decent working practices.

Amazon isn’t alone. The working conditions revealed at Sports Direct’s fulfilment center a few years ago were even worse. Amazon is responding to the current criticism with fancy videos of robots scooting around their warehouses and claim to have an impressive 100,000 robots. However, they still employ half a million people. Its really hard and expensive to get robots to pick up unfamiliar objects without dropping or squashing them. Elon Musk, struggling to get the output of his Tesla car plant up to 5000 cars a week, stepped back from attempting to robotise every stage, tweeting ‘humans are underestimated’. 
Amazon recently increased the pickers' wages by roughly 15%. This must be partly in response to the bad press and maybe a shortage of workers prepared to put up with the conditions, but a more cynical explanation is that, because Amazon makes most of its money from its cloud computing service, it is using the profit to attract staff to drive competitors out of business. 

Like most of my machines, the idea is a mix of things going on in the world and things going on in my life. Last winter, I spent a lot of time helping to clear out Rex Garrod’s mammoth workshop (I made The Secret Life of Machines TV series with Rex in the late 1980s but he sadly now has advanced dementia). Amongst other things I rescued were a mass of pneumatic rams and valves and some beautiful stainless steel chain. I set myself the goal to use some of his stuff in my next arcade machine. My first idea was to use a long exotic 'rodless' pneumatic ram and spent a few days making a rough prototype when we were cut off by snow drifts.

Rodless ram video 

I abandoned that when I got interested in the gig economy and had the idea of using Rex's chain for an Amazon warehouse. Unfortunately the beautiful chain was just too heavy to be practical. 

I then joined a couple of his big guided rams to make a walking platform. This seemed to have potential.

Pneumatic walking action video

Later I decided the pneumatic ram walking action was just too much effort, but the final machine still contains lots of Rex’s pneumatic fittings, valves and aluminium plate.


As usual the process of making it presented a succession of technical problems. Chain is about the most efficient form of transmission (less friction than gears or belts) but even so with such a long and tortuous path I was worried about friction and also about oil messing up the warehouse products. So worried in fact that I bought a really expensive ‘marathon’ chain for £400. This has oiled bushes in every link so shouldn’t need much oiling, and also has exceptionally low friction. Its lovely stuff, worth every penny.

One of the failed designs for the conductor rail  

The next problem was getting the little figure to walk and turn round. It was too tricky to do this mechanically so I had to get an electric circuit to the cart. Wherever I put the extra conductor ‘rail’, it snagged with the figure when in the process of turning round. I had almost completely abandoned the idea when I thought of using magnets on the cart to ‘stick’ to a steel rail. The chain itself was the return contact, but had more electrical resistance than I was expecting, so has to sit on brass guide strips. All this took about 6 weeks and at times it felt as if the machine was terminally bogged down.

The man and cart are made of Tufnol, an old fashioned textile reinforced plastic. Its one of my favourite materials. Very tough, strong and good for bearings - can be machined, sanded, and glued with Epoxy.  

Final walking and turning mechanism video

I don’t give up easily, so next tried a rough prototype of a less effortful walking mechanism (based on a park exercise machine). This worked first time. As I write though I’m still nervous whether the final ‘robust’ version will survive the energetic assault of the public.



foot mechanism video

The next problem was linking the chain movement to the player’s walking action. In past machines like Divorce I’ve used a direct mechanical linkage, with an electromagnetic clutch to disengage it. The ‘feedback’ feels intuitive and satisfying but players can damage the mechanisms when they are over energetic.

The test rig for the drive motor

The geared stepper motor

So for a change I decided to try ‘fly by wire’, translating the walking movement into chain movement electronically. This was influenced by meeting Dave, the nervous squirrel, an Arduino programming whizz, who offered his help. The first day’s experiments weren’t very encouraging. He bought an enormous stepper motor which made a terrible noise and had so much strength that it broke a key part of the man and cart. But we had a good time working together so we decided to persever. I bought a much smaller stepper motor on Ebay with a 5.1 planetary gearbox to increase the torque. Mounted on rubber bushes this solved the noise problem. When Dave returned everything sprang to life. He reprogrammed the Arduino using its built in tones function and now the motor feels very responsive.

The next job was to make the warehouse. Just making the shelving seemed a daunting task. Fortunately UCL’s Making Institute (a few blocks from Novelty Automation) allowed me to print them all in 3mm perspex on one of their laser cutters. Zoe even had some reject perspex that she donated for free! From youtube I knew the warehouse was a mix of boxed and unboxed items. Lizzie, who works for me at Novelty Automation, asked me if she could make the boxes. She came to my workshop and spent a day cutting and sanding loads of tiny tulipwood blocks and then took them home to paint.


I had no idea what to expect so I was thrilled when she showed me the finished boxes. I still wanted to add some ‘unboxed’ products and persuaded my friend Gary to come shopping. We spent a morning trawling Bexhill’s 28 charity shops (I didn’t know it but Charity shop tourism is quite a Bexhill speciallity). A surprising number of earrings and broaches look like miniature versions of bigger things. In the afternoon we went to a Dolls house shop in Steyning, north of Shoreham – a great day out.

Its a weird thing about all my machines that I have no idea how addictive any game will be until a machine is almost finished. I have little idea until I start programming the plc. I’m sure big arcade companies put enormous effort into the ‘game design’ before starting to build anything. My approach is hit or miss so some turn out better than others, but that's also true for all the big companies who do all their market research.

Once I’ve started programming the challenge is to make the experience as satisfying as possible. With the Fulfillment Center it felt satisfying changing direction and fun when the shortest route was in a counter intuitive direction. The products ‘popping up’ in your cart were also good ‘rewards’. However, the slower speeds felt more frustrating than I’d expected, so had to be limited as an occasional feature. Instead of picking three products which had been my original intention, I decided the game should be to try and pick all nine - though you get kicked off sooner if you’re too slow.

In the middle of programming my neighbour’s grandchildren, Iris and Nina, came to stay and they became my testers, telling me when the game was too easy or too hard. They even started making some of the larger objects for the shopping trolley. I greatly enjoyed their company.

Iris testing the machine video

Finally the time had come to try it on the pier. The move went well and over 40 people used it on its first day. But then of course it stopped working - there are always teething problems with a new machine. Generally though I’m cautiously optimistic, a couple of weeks after installation. The day after I wrote that a shaft of the walking action came loose, the little man snapped off his cart and the main stepper motor seized. Nothing terminal but took several days to sort out. A bigger problem was that about a third of people using the machine didn't get the idea of changing direction to pick the shortest route and immediately got fired. After a break working on other stuff for a couple of months, I returned to sort out the problems for half term. Simply revising the instructions made a big difference. People also started enjoying the walk to the final product (the gravestones). Here the cart slows down as you're told you've been working 10 hours and its hard to go quite so fast. Odd to have a slow climax, but it seems to work!   





about the site


links to
other sites

where to see stuff

contact me